Removal of emerging micropollutants from ...

Journal of Environmental Chemical Engineering 4 (2016) 1102–1109

Contents lists available at ScienceDirect

Journal of Environmental Chemical Engineering journal homepage: www.elsevier.com/locate/jece

Removal of emerging micropollutants from wastewater by activated carbon adsorption: Experimental study of different activated carbons and factors influencing the adsorption of micropollutants in wastewater R. Maillera,* ,1, J. Gasperia,* , Y. Coquetb , C. Deromea , A. Buletéc , E. Vullietc , A. Bressyd, G. Varraulta , G. Chebbod, V. Rochere a

LEESU (UMR MA 102, Université Paris-Est, AgroParisTech), Université Paris-Est Créteil, 61 avenue du Général de Gaulle, 94010 Créteil Cedex, France SAUR, Direction de la Recherche et du Développement, 1 rue Antoine Lavoisier, 78064 Guyancourt, France c Université de Lyon, Institut des Sciences Analytiques, UMR5280 CNRS, Université Lyon 1, ENS-Lyon, 5 rue de la Doua, Villeurbanne 69100, France d LEESU (UMR MA 102, Université Paris-Est, AgroParisTech), École des Ponts ParisTech, 6-8 avenue Blaise Pascal, Champs-sur-Marne, 77455 Marne-la-Vallée Cedex 2, France e SIAAP, Direction du Développement et de la Prospective, 82 avenue Kléber, 92700 Colombes, France b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 21 September 2015 Received in revised form 26 November 2015 Accepted 14 January 2016 Available online 16 January 2016

Activated carbon processes, initially designed for drinking water production, are tested for wastewater application in order to characterize their efficiency to remove micropollutants from wastewater treatment plants (WWTPs) discharges. In that purpose, a pilot was studied by the Paris sanitation service (SIAAP) and the water environment and urban systems laboratory (LEESU). The in-situ study raised several additional questions related to the structural and morphological properties of activated carbons, in order to select the proper material, the influence of operational parameters such as the activated carbon dose and the contact time, the role of organic matter concentration and composition, the presence of a residual concentration of methanol or the impact of ferric chloride addition. Thus, various complementary experiments were carried out at laboratory scale to improve the understanding of the micropollutants adsorption process on activated carbon, in particular on powdered activated carbon (PAC). The results have highlighted a strong link between the efficiency of PACs and their specific surface (BET), which can be easily estimated by their bulk density. The study of the sorption process has also confirmed the strong influence of the PAC dose and the rapidity of the sorption kinetic. From an operational point of view, the ferric chloride injection seems to slightly improve most of the detected compounds adsorption, probably thanks to the coagulation of the dissolved organic matter colloidal fraction. In contrary, the presence in the water of a residual concentration of methanol seems to have no impact on the micropollutant fate. The influence of the wastewater matrix is strong, with notably lower adsorption in water from primary settling compared to various WWTP discharges. However, the dissolved organic carbon concentration is not always sufficient to explain sorption competitions in wastewater, and the nature of the organic matter should be considered too. In particular, the carbon removal from biological treatments is the step that clearly modifies both the quantity and the composition of the organic matter. It has been observed that discharges from WWTPs operating with different biological processes (activated sludge, membrane bioreactor or biofiltration) have similar organic matter concentrations and compositions, and allows comparable removals of organic matter and micropollutants by adsorption. The lower performances on micropollutants observed in the settled water can be explained by the higher quantity of protein-like molecules (fluorophores Id and Ig), which compose the most competitive organic matter fraction for adsorption on activated carbon, compared to the other waters. ã 2016 Elsevier Ltd. All rights reserved.

Keywords: Adsorption Organic matter Pharmaceuticals Activated carbon Wastewater

* Corresponding authors. E-mail addresses: [email protected] (R. Mailler), [email protected] (J. Gasperi). Present address: SIAAP, Direction du Développement et de la Prospective, 82 avenue Kléber, 92700 Colombes, France.

1

http://dx.doi.org/10.1016/j.jece.2016.01.018 2213-3437/ ã 2016 Elsevier Ltd. All rights reserved.

R. Mailler et al. / Journal of Environmental Chemical Engineering 4 (2016) 1102–1109

1. Introduction The presence of a large range of emerging micropollutants, particularly pharmaceuticals and hormones, personal care products or pesticides, have been highlighted in wastewater treatment plant (WWTP) discharges [1–4]. Even if several hydrophobic, volatile or biodegradable compounds are strongly removed by conventional wastewater treatments [5,6], most of micropollutants are poorly removed. Thus, various strategies of contamination reduction, such as source reduction, WWTP optimization or tertiary treatment implementation, are assessed by the scientific community and water treatment engineers. Among them, the implementation of tertiary treatments dedicated to micropollutants elimination represents a relevant solution. In particular, adsorption on activated carbon appears to be efficient, adapted to any types of WWTP and relatively cheap [7], together with not producing oxidation by-products. In this context, the Parisian public sanitation service (SIAAP) and the water environment and urban systems laboratory (LEESU) study the CarboPlus1 process, in collaboration with SAUR teams. This process is based on a fluidized bed of activated carbon (powder – PAC - or micro-grain – mGAC) to remove micropollutants by adsorption. A large-scale pilot based on this technology was set up at the Seine Centre (SEC) WWTP (240,000 m3/day) to characterize the efficiency of activated carbon to remove a wide range of pollutants from WWTP discharges. The in-situ results are presented in Mailler et al. for PAC and Mailler et al. for mGAC [4,8]. In parallel to the in situ study, complementary laboratory scale experiments were conducted to (i) improve the understanding of the micropollutant adsorption mechanisms in wastewater application, in particular with PAC, and (ii) better understand the relationships between activated carbon properties and the adsorption processes in wastewater. Activated carbon is characterized by different structural and morphological properties that can affect adsorption [9–12]. The understanding of the links between activated carbon properties and pollutant adsorption is still incomplete. Thus, the first axis consists in studying the relationships between activated carbon properties and their efficiency to remove micropollutants. In particular, a focus is performed on the specific BET (Brunauer, Emmett and Teller) surface and the bulk density. The second axis consists in studying the adsorption mechanism in wastewater representing a complex matrix. Indeed, the literature highlighted that the matrix where the adsorption is performed plays a crucial role in the fate of molecules [13–17], particularly the organic matter which competes with pollutants for adsorption through direct sites competition or pore blocking. The adsorption interactions were already deeply studied in the literature, but most of the studies were conducted for drinking water production or in surface waters, and these mechanisms were poorly assessed in wastewater [18,19]. In this context, the influence of both the quantity and the composition of the organic matter on micropollutant adsorption in wastewater was assessed using 3D spectrofluorometry. The study of such interactions between activated carbon, emerging micropollutants and organic matter in wastewater using this innovative technique and adopting field conditions (limited contact time with PAC) is novel. The novelty of this study is also the assessment of different types (settled, carbon treatment effluent, various WWTP effluents) and quality of wastewater, as well as the effects of (i) the FeCl3 injection and (ii) the presence of a residual methanol concentration in the effluents on the adsorption of emerging micropollutants. Such influences were not or poorly documented in the literature [20]. This article summarizes the results from the laboratory scale experiments conducted within this project. First, the relationships between micropollutant removals and activated carbon properties

1103

are assessed. Then, the influence of the dose of carbon, the contact time, the organic matter quantity and composition, the presence of a residual concentration of methanol and the presence of FeCl3 were studied. 2. Materials and methods 2.1. Activated carbon characterization 11 adsorbents, including micro and mesoporous PACs and

mGAC, have been selected based on their technical datasheets provided by the producers. Among them, 3 are commercialized by DaCarb1 (PB 1701, PB 170-4001 and PC 10001 —France), 3 by Chemviron1 (WP 2351, Carbsorb 281 and Cyclecarb 3051— Belgium), 2 by Norit1 (W 351 and SA Super1—Netherlands) and 3 by Jacobi1 (LP 391, MP 251 and Hydro XP 171—Sweden). These activated carbons are recognized for their high affinity for organic pollutants, but their use in wastewater was poorly studied in the literature. Several structural properties such as the specific BET surface (m2/g), the porous volume (mL/g), the pore size distribution and the bulk density (g/cm3) have been measured on the 11 activated carbons. In addition, the particle size distribution and the micropollutant elimination have been determined for 4 of them: PB 1701, WP 2351, W 351 and PC 10001. The specific BET surface and the pore size distribution were measured with an ASAP 2010 Micromeritics analyzer equipped with a degasing station and a gas isotherm adsorption analyses station (nitrogen), according to the conventional methods used to determine these parameters. Fresh activated carbon samples (100 mg) were degased at least 12 h before analysis. Results correspond to the specific surface in m2/g obtained with the BET method. The pore size distribution is determined with the BJH (Barrett, Joyner and Halenda) method, using the desorption curve of the same gas on the same analyzer. This method allows also calculating the micro and mesoporous volumes. The bulk density was measured by weighting, with a high precision balance (0.01 mg), 50 to 100 mL of activated carbon, measured with a 100 mL graduated cylinder. The activated carbon is introduced by doses (10 mg) and is compacted by patting gently the cylinder several times every 10 mL to minimize the vacuum between particles. The particles size distribution of the 4 PACs was measured with a Mastersizer 2000 Malvern laser particle sizer. Every analysis corresponds to 15,000 light diffraction measures. 3 scans were performed at least per sample. 2.2. Pollutants and analytical procedures Samples were filtered on 0.7 mm glass fibers filters (GF/F) before all analyses. For every sample, several parameters were measured in the dissolved phase: UV absorbance (Secomam1 UV Light) at 254 nm (UV-254), dissolved organic carbon (DOC) and concentrations of 17 pharmaceuticals and 2 pesticides (list and limit of quantification – LQ – in Supporting material—Table S1). Analytical protocols are validated and given in [4]. 3D fluorescence spectrometry analyses were also performed after filtration. This method is used to characterize the dissolved organic matter (DOM) present in the sample. It gives information about the nature and origin of the DOM through an excitation emission matrix (EEM) containing all excitation and emission spectra obtained for the DOM. Indeed, the different fractions of the DOM fluoresce in different zones of the spectrum. The detailed description of 3D fluorescence spectrometry is given in [21]. Analyses have been performed with a Jasco FP-8300 spectrofluorometer equipped with a 1 cm quartz cell. The ranges of

1104


wavelengths were 240–450 nm in excitation (intervals of 2 nm) and 300–600 nm in emission (intervals of 5 nm). The EEM obtained was treated following the fluorophores defined in Parlanti et al. [22], as indicated for SEC settled water and discharges in Supporting material—Fig. S1. Ia and Ia’ correspond to humic-like substances, Ib results from a recent autochthone material production, Ig and Id respectively represents tyrosine-like and tryptophan-like proteins. Then, Ia/Ia’, Ib/Ia’, Id/Ia’ and Ig/Ia’ indexes were calculated to evaluate the distribution of DOM. Finally, the humification index (HIX) and the biological activity index (BIX) were also evaluated [22,23]. HIX is the ratio H/L of two spectral areas (H = maximum between 300 and 345 nm; L = maximum between 435 and 480 nm) from the emission spectrum scanned at an excitation wavelength of 254 nm. BIX is the ratio of the intensity at an emission wavelength of 380 nm and the intensity at 430 nm, scanned at an excitation wavelength of 310 nm.

contacting individually 10 mg/L of each PAC with 1 L of SEC WWTP discharges during 45 min under strong mixing. 45 min is representative of the contact time in PAC tertiary processes [4]. To evaluate PACs, UV-254, DOC and micropollutants were measured both before and after PAC contact. 2.3.2. PAC dose and adsorption kinetic Regarding its performances on micropollutant removals (results presented after), PB 1701 was selected to characterize the relationship between PAC dose, contact time and micropollutant removal. First, 10 mg/L of this PAC were contacted with 1 L of SEC discharges under strong mixing during different contact times (5–10–30–45–60 min). Then, 3 PAC doses (5–10–20 mg/L) were tested with 1 L of SEC discharges and during 45 min. UV-254, DOC and micropollutants were measured both before and after PAC contact.

2.3. Laboratory experiments protocols The laboratory experiments have been performed in the SEC WWTP (Colombes, France) between April 2013 and May 2015. This water is characterized by a relatively low content of dissolved organic matter, with an average DOC value of 5.6 0.9 mgC/L [4]. Wastewater was punctually sampled from the WWTP discharges with 10 L glass bottles, properly rinsed and grilled, and used the same day for experiments (2 h after sampling maximum). In the different experiments, the water was contacted with PAC under strong mixing in rinsed and grilled (500 C) 2 L batches covered with aluminum paper. Then, samples were filtered on 0.7 mm GF/F filters after experiments and conditioned before analyses. UV254 was measured directly after filtration, while DOC was analyzed within 24 h and micropollutants were analyzed within 48 h (storage at 4 C). 2.3.1. Efficiency comparison of 4 PACs The efficiency of 4 PACs (PB 1701, WP 2351, W 351 and PC 10001—Table 1) to remove micropollutants was assessed, by

2.3.3. DOM and adsorption To determine the role of both the quantity and the quality of the DOM in the adsorption process, 7 types of wastewater featuring by different levels of organic matter have been contacted under strong mixing with PAC (10 mg/L of PB 1701 during—45 min). These wastewaters have similar concentrations concerning the studied micropollutants since these WWTPs receive water from the same catchment (Paris conurbation). Among the tested waters, 4 were WWTP discharges, from SEC, Seine Aval (SAV), Seine Morée (SEM) and Seine Amont (SAM) WWTPs. The WWTPs layouts are given in Supporting material—Fig. S2. The 3 other tested waters were outlet waters from a physico-chemical lamellar settling unit (SEC settled water), a carbon biofiltration unit (SEC carbon biofiltration stage), and the SEC settled water diluted with ultra-pure water to reach a DOC level comparable to the WWTP discharges. The layouts of the different studied WWTPs are given in Supporting material— Table S1, as well as the sampling points (red circles). UV-254, DOC and 3D fluorescence spectrometry have been measured both before and after PAC contact for the 7 wastewaters, while the

Table 1 Characterization and performances of the 4 PACs studied.

Producer Raw material Bulk density (g/cm3) Specific BET surface (m2/g) Pore size distribution (%) (micro, meso, macroporous)a Pore volume (micro + meso) (mL/g) Particle size distribution (mm) d10–d50–d90 Concentration in SEC WWTP discharges (ng/L) Atenolol Atrazine Carbamazepine Ciprofloxacin Diclofenac Diuron Erythromycin Ketoprofen Metronidazole Norfloxacin Ofloxacin Propranolol Roxithromycin Sulfamethoxazole Trimethoprim Average removal of investigated pollutants

435 6 267 425 303 57 585 61 33 197 1632 206 68 186 355

PB 1701

WP 2351

W 351

PC 10001

DaCarb Wood 0.30 957 28 54%–35%–11%

Chemviron Coal 0.38 909 30 53%–31%–16%

Norit Peat 0.33 768 19 45%–45%–10%

DaCarb Coconut 0.54 458 14 59%–29%–12%

0.5066 3.4–16.2–58.9

0.4841 3.4–18.0–58.0

0.4876 3.2–19.4–86.2

0.2435 3.7–31.7–129.4

Micropollutant removal (%) with 10 mg/L of PAC during 45 min of contact 54 44 41 40 35 37 63 44 48 77 63 62 32 22 28 75 66 60 26 17 22 17 22 31 16 13 10 79 69 62 79 71 68 81 76 70 43 38 40 51 60 57 46 36 33 52 45 45

21 12 16 49 5 24 5 10 0 58 46 46 21 60 10 26

a Pore size distribution (%): micro < 2 nm, meso 2–50 nm and macropores > 50 nm. Bezafibrate, lorazepam, naproxen and oxazepam were not detected during this experiment.


micropollutants have been assessed only with SEC settled water, SEC carbon biofiltration effluent, SEC and SAV discharges. 2.3.4. Influence of the residual methanol concentration in the WWTP discharges A residual methanol concentration is present in both SEC and SAV discharges (20% of DOC, SIAAP source) because of the postdenitrification stage. Indeed, denitrifying microorganisms needs a carbon source and the carbon content is too low as denitrification is performed in post-denitrification configuration in both SEC and SAV [24]. Methanol is then added during this stage as carbon source, but its consumption by the microorganisms is not complete. Operationally speaking, it is interesting to study the impact of this residual concentration on the micropollutant adsorption by activated carbon. Thus, the performances obtained with the SEC discharges have been compared to performances obtained with the same water spiked with methanol (6 and 12 mgC/L; methanol analytical grade Merck1). The previously cited micropollutants, as well as UV-254 and DOC, were measured before and after contact with PAC (45 min). The adsorbability of methanol has also been studied by following DOC and preliminary tests demonstrated that no adsorption of methanol occurs when contacting 12 mgC/L of methanol in ultra-pure water with 10 mgPAC/L. 2.3.5. Influence of the ferric chloride The CarboPlus1 process operating in PAC configuration requires the injection of FeCl3 to stabilize the PAC bed and prevent any leakage. This substance is known to destabilize particles by neutralization of the surface charges [25], which could impact the adsorption of micropollutants. To evaluate the impact of FeCl3, 1 L of SEC discharges have been contacted with PAC alone (10 mg/L of PB 1701), FeCl3 alone (2.5 gFeCl3/m3) and both PAC (10 mg/L of PB 1701) and FeCl3 (2.5 gFeCl3/m3) under strong mixing during 45 min. The previously cited micropollutants, as well as UV254 and DOC, were measured before and after contact with PAC. The FeCl3 product (commercial solution, 40% FeCl3–Quaron SA1) and dose were the same as in the pilot.

1105

investigated micropollutants is 52%, 45%, 45% and 26% respectively for PB 1701, WP 2351, W 351 and PC 10001. The micropollutant removal is well correlated with the specific BET surface (Fig. 1): the higher the specific BET surface, the higher the micropollutant removal. This correlation is significant (p-value